search

hpcaitech / Open-Sora

Open-Sora: Democratizing Efficient Video Production for All
https://hpcaitech.github.io/Open-Sora/
Apache License 2.0
21.48k stars 2.06k forks source link

双卡运行 Gradio Demo 的时候好像只有一个显卡在推理 #376

Closed ustcxmwu closed 3 months ago

ustcxmwu commented 3 months ago

运行命令 CUDA_VISIBLE_DEVICES=0,1 python gradio/app.py

推理时报显存不够 但是显存显示 ` nvidia-smi Tue May 7 10:29:12 2024
+-----------------------------------------------------------------------------------------+ | NVIDIA-SMI 550.54.15 Driver Version: 550.54.15 CUDA Version: 12.4 | |-----------------------------------------+------------------------+----------------------+ | GPU Name Persistence-M | Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap | Memory-Usage | GPU-Util Compute M. | | | | MIG M. | |=========================================+========================+======================| | 0 NVIDIA GeForce RTX 4090 Off | 00000000:3B:00.0 Off | Off | | 44% 29C P8 18W / 450W | 24183MiB / 24564MiB | 0% Default | | | | N/A | +-----------------------------------------+------------------------+----------------------+ | 1 NVIDIA GeForce RTX 4090 Off | 00000000:AF:00.0 Off | Off | | 44% 28C P8 22W / 450W | 14MiB / 24564MiB | 0% Default | | | | N/A | +-----------------------------------------+------------------------+----------------------+

+-----------------------------------------------------------------------------------------+ | Processes: | | GPU GI CI PID Type Process name GPU Memory | | ID ID Usage | |=========================================================================================| | 0 N/A N/A 3547 G /usr/lib/xorg/Xorg 4MiB | | 0 N/A N/A 72296 C python 24166MiB | | 1 N/A N/A 3547 G /usr/lib/xorg/Xorg 4MiB | +-----------------------------------------------------------------------------------------+ ` 有类似的情况吗

zhengzangw commented 3 months ago

For the v1.1, we have not yet completed the multiple cards inference, and thus can only support one GPU. We are working on supporting sequence parallelism in inference.

ustcxmwu commented 3 months ago

For the v1.1, we have not yet completed the multiple cards inference, and thus can only support one GPU. We are working on supporting sequence parallelism in inference.

thanks